summaryrefslogtreecommitdiffstats
path: root/llvm/lib/Transforms/InstCombine/InstCombineVectorOps.cpp
diff options
context:
space:
mode:
authorBob Wilson <bob.wilson@apple.com>2010-10-29 22:02:50 +0000
committerBob Wilson <bob.wilson@apple.com>2010-10-29 22:02:50 +0000
commitcb11b48e7a9d438defa72ed96001ec3f26509378 (patch)
tree5ba3854f33a9aab9a774bf88556f4c53d4665fea /llvm/lib/Transforms/InstCombine/InstCombineVectorOps.cpp
parent31f23b48ba4003f78ddad42a8ccc544b64d5f08d (diff)
downloadbcm5719-llvm-cb11b48e7a9d438defa72ed96001ec3f26509378.tar.gz
bcm5719-llvm-cb11b48e7a9d438defa72ed96001ec3f26509378.zip
Make instcombine a little more aggressive in combining vector shuffles.
Allow splats even if they don't match either of the original shuffles, possibly due to undef entries in the shuffles masks. Radar 8597790. Also fix some 80-column violations. llvm-svn: 117719
Diffstat (limited to 'llvm/lib/Transforms/InstCombine/InstCombineVectorOps.cpp')
-rw-r--r--llvm/lib/Transforms/InstCombine/InstCombineVectorOps.cpp37
1 files changed, 22 insertions, 15 deletions
diff --git a/llvm/lib/Transforms/InstCombine/InstCombineVectorOps.cpp b/llvm/lib/Transforms/InstCombine/InstCombineVectorOps.cpp
index a58124d7032..634add86a81 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombineVectorOps.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineVectorOps.cpp
@@ -515,37 +515,44 @@ Instruction *InstCombiner::visitShuffleVectorInst(ShuffleVectorInst &SVI) {
// we are absolutely afraid of producing a shuffle mask not in the input
// program, because the code gen may not be smart enough to turn a merged
// shuffle into two specific shuffles: it may produce worse code. As such,
- // we only merge two shuffles if the result is one of the two input shuffle
- // masks. In this case, merging the shuffles just removes one instruction,
- // which we know is safe. This is good for things like turning:
- // (splat(splat)) -> splat.
+ // we only merge two shuffles if the result is either a splat or one of the
+ // two input shuffle masks. In this case, merging the shuffles just removes
+ // one instruction, which we know is safe. This is good for things like
+ // turning: (splat(splat)) -> splat.
if (ShuffleVectorInst *LHSSVI = dyn_cast<ShuffleVectorInst>(LHS)) {
if (isa<UndefValue>(RHS)) {
std::vector<unsigned> LHSMask = getShuffleMask(LHSSVI);
if (LHSMask.size() == Mask.size()) {
std::vector<unsigned> NewMask;
- for (unsigned i = 0, e = Mask.size(); i != e; ++i)
- if (Mask[i] >= e)
- NewMask.push_back(2*e);
- else
- NewMask.push_back(LHSMask[Mask[i]]);
+ bool isSplat = true;
+ unsigned SplatElt = 2 * Mask.size(); // undef
+ for (unsigned i = 0, e = Mask.size(); i != e; ++i) {
+ unsigned MaskElt = 2 * e; // undef
+ if (Mask[i] < e)
+ MaskElt = LHSMask[Mask[i]];
+ // Check if this could still be a splat.
+ if (MaskElt < 2*e) {
+ if (SplatElt < 2*e && SplatElt != MaskElt)
+ isSplat = false;
+ SplatElt = MaskElt;
+ }
+ NewMask.push_back(MaskElt);
+ }
// If the result mask is equal to the src shuffle or this
// shuffle mask, do the replacement.
- if (NewMask == LHSMask || NewMask == Mask) {
+ if (isSplat || NewMask == LHSMask || NewMask == Mask) {
unsigned LHSInNElts =
cast<VectorType>(LHSSVI->getOperand(0)->getType())->
getNumElements();
std::vector<Constant*> Elts;
+ const Type *Int32Ty = Type::getInt32Ty(SVI.getContext());
for (unsigned i = 0, e = NewMask.size(); i != e; ++i) {
if (NewMask[i] >= LHSInNElts*2) {
- Elts.push_back(UndefValue::get(
- Type::getInt32Ty(SVI.getContext())));
+ Elts.push_back(UndefValue::get(Int32Ty));
} else {
- Elts.push_back(ConstantInt::get(
- Type::getInt32Ty(SVI.getContext()),
- NewMask[i]));
+ Elts.push_back(ConstantInt::get(Int32Ty, NewMask[i]));
}
}
return new ShuffleVectorInst(LHSSVI->getOperand(0),
OpenPOWER on IntegriCloud